Dear Dr Haddidi,
I think your question desserves a better description especially for people like me who are not used with your specific area of work but are interested in diagnostic test accuracy study.
My first answer would be a "yes", if you have the nomogram with at least 2 specific examples and pretest probability you could find the LR based on post-test probability and Bayes rule. You could also alternatively test the specific test in another population to compare with the accuracy in your field setting. This would help you to then report the test results and outcomes observed.
Calculation of the ROC curve would then naturally follow using any specific software (ex: R, SAS or Stata...)
Thank you very much Mr Buzinsky. I give all my data in excel to a competent epidemiologist. He use the SPSS, and than he calculate the cut off, the logiciel draw the ROC curve and calculate the AUC also. And the intrinseque valuesi were good.
Its certainly possible to work out all those measures by tabulating the true node status against a binary output from the MSKCC ie scoring as above or below a cutoff - those measures are row or column percentages. The ROC curve itself plots sensitivity and 1-specificity. Did you want to assess other measures of predictive ability such as calibration as discussed here https://www.youtube.com/watch?v=v_e85TDYj9Y
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Physiological Processes
- Stress